In ovo feeding of nutrients and its impact on post-hatching water and feed deprivation up to 48 hr, energy status and jejunal morphology of chicks using response surface models.

A Box-Behnken design (BBD) in a response surface methodology (RSM) was used to investigate the response of broiler chicks to in ovo feeding (IOF) of beta-hydroxy beta-methylbutyrate (HMB), dextrin and the timing of the first water and feed deprivation. On day 18th of incubation, 1,500 eggs were randomly assigned to 15 experimental runs of BBD, each with 4 replicates, as 3 levels IOF of HMB (0%, 0.5% and 1%) and dextrin (0%, 20% and 40%), and 3 levels of the first water and feed deprivation (6, 27 and 48 hr). Day-old chicks from each replicate were then used to assess the effect of IOF and time first water and feed access on chick's responses. The IOF of dextrin leads to respectively 9.7%-15.5% lower hatchability for 20% and 40% inclusion (p < .05), whereas HMB inclusion appeared with no effect on hatchability (p > .05). Administration of dextrin or HMB into the amnion of embryos elevated length, width and surface area of villus, and increased glycogen content of liver and breast (p < .05). In all parameter models, the linear terms showed highest contribution (R2  = 0.81-0.97) to explain existing variation in chick's responses. The first water and feed deprivation had largest effect on BW2 and glycogen content of liver and breast. It is concluded that if possible, place chicks before 7 hr of hatch to preserve BW loss and have maximum response from IOF. If not possible, use IOF with 40% dextrin + 0.5% HMB to preserve gut integrity and energy status up to 48 hr. This should give advantage to chicks to recover fast after feeding, but that would have to be confirmed by trials growing birds to slaughter age.

Alternative growth functions for predicting body, carcass, and breast weight in ducks: Lomolino equation and extreme value function.

In this study, 2 alternative growth functions, the Lomolino and the extreme value function (EVF), are introduced and their ability to predict body, carcass, and breast weight in ducks evaluated. A comparative study was carried out of these equations with standard growth functions: Gompertz, exponential, Richards, and generalized Michaelis-Menten. Goodness of fit of the functions was evaluated using R(2), mean square error, Akaike information criterion, and Bayesian information criterion, whereas bias factor, accuracy factor, Durbin-Watson statistic, and number of runs of sign were the criteria used for analysis of residuals. Results showed that predictive performance of all functions was acceptable, though the Richards and exponential equations failed to converge in a few cases for both male and female ducks. Based on goodness-of-fit statistics, the Richards, Gompertz, and EVF were the best equations whereas the worst fits to the data were obtained with the exponential. Analysis of residuals indicated that, for the different traits investigated, the least biased and the most accurate equations were the Gompertz, EVF, Richards, and generalized Michaelis-Menten, whereas the exponential was the most biased and least accurate. Based on the Durbin-Watson statistic, all models generally behaved well and only the exponential showed evidence of autocorrelation for all 3 traits investigated. Results showed that with all functions, estimated final weights of males were higher than females for the body, carcass, and breast weight profiles. The alternative functions introduced here have desirable advantages including flexibility and a low number of parameters. However, because this is probably the first study to apply these functions to predict growth patterns in poultry or other animals, further analysis of these new models is suggested.

Narushin-Takma models as flexible alternatives for describing economic traits in broiler breeder flocks.

Three Narushin-Takma (NT) models (NT1, NT2, and NT3) were examined for their ability to describe different curves obtained from broiler breeder flocks. The models NT1, NT2, and NT3 comprise 3 flexible mathematical functions (rational polynomial functions) with 5, 6, and 7 parameters, respectively. The characteristics fitted were BW, egg production, egg mass, egg weight, first- and second-grade eggs, hatchability, feed intake, and feed conversion ratio. To evaluate the ability of these NT models to fit the different curves, comparisons were made with more commonly fitted functions (Gompertz, modified compartmental, Richards, Adams-Bell, and Lokhorst). Comparisons revealed a higher accuracy of fit with the NT models, proving their general flexibility. This study likely represents the first time a generic model has been demonstrated to fit all these characteristics satisfactorily. Results showed that in most cases, NT3, because of its greater number of parameters, gave the highest accuracy of prediction. The NT models are likely to fit most curves and are therefore advocated for accurate prediction of other traits with a minimum of mathematical complexity.

Predicting caloric and feed efficiency in turkeys using the group method of data handling-type neural networks.

Neural networks (NN) are a relatively new option to model growth in animal production systems. One self-organizing submodel of artificial NN is the group method of data handling (GMDH)-type NN. The use of such self-organizing networks has led to successful application of the GMDH algorithm over a broad range of areas in engineering, science, and economics. The present study aimed to apply the GMDH-type NN to predict caloric efficiency (CE, g of gain/kcal of caloric intake) and feed efficiency (FE, kg of gain/kg of feed intake) in tom and hen turkeys fed diets containing different energy and amino acid levels. Involved effective input parameters in prediction of CE and FE were age, dietary ME, CP, Met, and Lys. Quantitative examination of the goodness of fit for the predictive models was made using R2 and error measurement indices commonly used to evaluate forecasting models. Statistical performance of the developed GMDH-type NN models revealed close agreement between observed and predicted values of CE and FE. In conclusion, using such powerful models can enhance our ability to predict economic traits, make precise prediction of nutrition requirements, and achieve optimal performance in poultry production.

Prediction model for true metabolizable energy of feather meal and poultry offal meal using group method of data handling-type neural network.

A group method of data handling-type neural network (GMDH-type NN) with an evolutionary method of genetic algorithm was used to predict the TME(n) of feather meal (FM) and poultry offal meal (POM) based on their CP, ether extract, and ash content. Thirty-seven data lines consisting of 15 FM and 22 POM samples were collected from literature and used to train a GMDH-type NN model. A genetic algorithm was deployed to design the whole architecture of the GMDH-type NN. The accuracy of the model was examined by R(2) value, adjusted R(2), mean square error, residual standard deviation, mean absolute percentage error, and bias. The developed model could accurately predict the TME(n) of FM or POM samples from their chemical composition. The R(2) for the GMDH-type NN model had a higher accuracy of prediction than 2 models reported previously. This study revealed that the novel modeling of GMDH-type NN with method of genetic algorithm can be used to predict the TME(n) of poultry by-products.

Predicting performance of broiler chickens from dietary nutrients using group method of data handling-type neural networks.

1. Successful artificial neural network (ANN) applications have been found for many areas. One sub-model of ANNs is the group method of data handling-type neural networks (GMDH-type NNs). The use of self-organising networks leads to successful application in a broad range of areas. However, the use of such methods is not common in poultry science. 2. Broiler chicken nutrition is recognised as a biological system consisting of a complex set of interconnected variables. The adequate information on nutrients (variables), such as metabolisable energy (ME) and amino acid requirements, can help to establish specific feeding programmes, defining optimal performance and reducing production costs. 3. This study addressed the question of whether GMDH-type NNs can be used to estimate the performance of broiler chickens (output) based on specified variables-inputs (dietary crude protein (CP), ME, ME/CP, methionine (Met), lysine (Lys), ME/Met and ME/Lys)-for a commercial broiler chicken farm. The recorded data from 10 broiler chicken flocks were obtained, from March 2003 to April 2005, corresponding to 52 data lines. 4. The results suggested that the GMDH-type NNs may provide an effective means of recognising the patterns in data and accurately predicting the performance of broiler chickens based on investigating inputs. In addition the polynomial equations obtained can be used to optimise the performance of broilers.

Hyperbolastic models as a new powerful tool to describe broiler growth kinetics.

The mathematical models for describing growth kinetics are important tools to examine biological parameters, such as BW at specific time, maximum growth response, and growth rates. Classical growth models such as Gompertz and Richards have been extensively used in broiler studies. To obtain a more intuitive understanding of growth, a class of flexible growth models containing 3 and 4 parameters were developed as hyperbolastic models. These models provide description of growth behaviors for continuous output in different fields, for instance cancer and stem cell growth. This study was conducted to find out if 3 flexible new hyperbolastic growth models, called type 1, 2, and 3 (H1, H2, and H3), may be used to illustrate relationship between live weight and age in broilers and also how effective such models would be compared with 2 classical growth models, namely Gompertz and Richards. A set of growth data over 70 d, obtained from 18 male broilers, was used to fit growth models. Goodness of fit of the models was determined using mean square error, R(2), and residual standard deviation. It was revealed that the 3 new models may be used to fit broiler growth data successfully and could be implanted in SAS PROC NLIN. Goodness of fit criteria refers better fit with H3, presumably due to its greater flexibility, followed by the Richards, Gompertz, H2, and H1. In conclusion, it seems that the H3 can be considered as a more useful tool for modeling broiler growth. However, it is reasonable to compare models for fit before selection of the more accurate one.